[Syl20]

Hi,
For the ZAR users.
This evaluation seems interesting but i still have uncertainties...
Thanks to convice me on the following point: when responder as a fit and Ace of trumps, this Ace counts 8 Zars : 4Hcp + 2Cont. + 2Hon. in trumps suit ?
Seems a lot, no ?
How many Zars do you count on:
♠KQJxx
♥AQx
♦Ax
♣xxx
opposite:
♠Axx
♥xxx
♦Qxxxx
♣xx

On Zar's site. It is claimed that 99% of the games made on a championship were found with Zar evaluation.
This looks impressive but what would better suit me is the ratio:
"number of game bid with ZAR and not by conventional evaluation AND made" to
"number of game bid with ZAR and not by conventional evaluation AND off".

Any comments welcome.

[Free]

I'm not a zar user, so I won't respond to this first question. However, I've seen many good evaluation statistics but indeed, never one where zar fails and the field scores. I've also noticed that many slams bid by zar are on a finesse! So I'm still not that impressed to switch over... I'm also interested in these 'negative' stats, if they exist

[skorchev]

Quote

On Zar's site.

What is Zar's site?


Stefan

[inquiry]

skorchev, on Sep 17 2004, 06:24 AM, said:

Quote

On Zar's site.

What is Zar's site?


Stefan

Click here for Zar's site

Ben

[inquiry]

Syl20, on Sep 17 2004, 04:41 AM, said:

Hi,
For the ZAR users.
This evaluation seems interesting but i still have uncertainties...
Thanks to convice me on the following point: when responder as a fit and Ace of trumps, this Ace counts 8 Zars : 4Hcp + 2Cont. + 2Hon. in trumps suit ?
Seems a lot, no ?
How many Zars do you count on:
♠KQJxx
♥AQx
♦Ax
♣xxx
opposite:
♠Axx
♥xxx
♦Qxxxx
♣xx

On Zar's site. It is claimed that 99% of the games made on a championship were found with Zar evaluation.
This looks impressive but what would better suit me is the ratio:
"number of game bid with ZAR and not by conventional evaluation AND made" to
"number of game bid with ZAR and not by conventional evaluation AND off".

Any comments welcome.

The ACE in your partner's suit counts as 7 points, not 8. You get 4 for high card value, 2 for control, but only one for fitting honor in partner suit.

The north hand you show has:
11 distributional zar ponts, 16 hcp, 5 control points = 32 Zar points

The South hand you show has:
11 distributional, 6 hcp, 2 for control for 19 Zar points.

When north shows spades, south gets to add one pont for the fitting heart honor.

This totals 52 (32+19+1 fit). In theory, with fit enough for game. Not all 52 Zar point hands will make game, and you will not be able to tell with certainty if you have 52 during bidding anyway.

Additional comments. Bidding ZAR points is aggressive. You will bid a lot more slams and games, and as a result you will go down in games more often (but you will make a lot more too). One way to keep from overbidding is to use negative zar points as well. Take off for horrible distribution, take off points for shortness in trumps. I find Zar ponts quite effective if used with COMMON sense too.

Thsi forum has a lot of theads on zar points and other evaluation schemes. You may want to read the following threads (ZAR himself particpates here). Below is only a partial listing of some useful thread on ZAR points

Zar points, useful or waste of energy, New to the concept, does it help...

Fine tuning ZAR points., Help me with my adjustments...

Zar evaluation and 2004 Cavendish, Invitational Pairs

Zar points application in competitive bidding

ZAR Distribution Points, An easier way to calculate

Zar points and hands with no defensive strength, Problems if opps compete

[Syl20]

inquiry, on Sep 17 2004, 07:49 AM, said:

The ACE in your partner's suit counts as 7 points, not 8. You get 4 for high card value, 2 for control, but only one for fitting honor in partner suit.

Additional comments. Bidding ZAR points is aggressive. You will bid a lot more slams and games, and as a result you will go down in games more often (but you will make a lot more too). One way to keep from overbidding is to use negative zar points as well. Take off for horrible distribution, take off points for shortness in trumps.  I find Zar ponts quite effective if used with COMMON sense too.

Thanks for the answer.
I made a tool to calculate Zar points (thanks Free for the advice ) with positive corrections for trump honnours, trump length if doubletons and singletons and negative corrections in case of singleton or doubleton Queen and Jacks and singleton Kings.
I asked "bridgelab.exe" to make 18000 boards with an opening 1♠ (5+♠, 12-17Hcp) and a fit for responder (5+Hcp, 3+♠).
Then, i calculated the number of tricks in ♠ for every board (with "gib.exe") and classified the number of tricks for various combined Zar points.
Although, i neither consider any opponents bidding (for other corrections) nor double fits, i found that the average tricks for 52 Zars is 9.0x and you need 56 Zars to get 10 tricks on average. I assume that we wish at least a probability of 50% to make the contract to be happy with the estimator.
Any comments ?

[inquiry]

Syl20, on Sep 22 2004, 11:46 AM, said:

Then, i calculated the number of tricks in ♠ for every board (with "gib.exe") and classified the number of tricks for various combined Zar points.
I found that the average tricks for 52 Zars is 9.0x and you need 56 Zars to get 10 tricks on average. I assume that we wish at least a probability of 50% to make the contract to be happy with the estimator.
Any comments ?

Well, I believe your approach is sound way to go. However, I wonder if your corrections and counting was accurate. People make a lot of mistakes in trying to interpret zar points. For instance, for the longest time, I counted 1 point for each honor in "our suit" was a fit was found from both sides of the table.. Thus, with

KQxx opposite AJxx

I would count 10 hcp, 3 controls, and 4 point for "fitting honors" for a total of 17 ZAR POINTS. This is not correct. Whoever bids this suit first, can not count the two extra points for fitting honors. So this is only 15 Zar fit points.

There are other gotcha's as well. Some have tripped me up, some have tripped up others. A big one is people count way to much for shortness once a fit is found... New users often fail to grasp when you can, and when you can not count the short suit points. So if you are going to make the claim that you need 56 zar points for game, it is imporant that you have caluculated the points correctly. ght. So, could you either show 10 pair of hands and tell us what you think the zar count is, or better yet, explain in detail your algorith for calculating

1) Initial Zar points (I am sure you got this part right)
2) Fit Zar points
3) Negative Zar points.

My own research suggest the 9.0 tricks for 52 zar points is too low. It is less than 10, but not nearly that much, which suggest that you are probably calculating the points wrong (or I am).

Ben

[ArcLight]

What is bridgelab.exe?

I looked for it on the web and couldn't find it.
I would find it useful if there was some software I could use to generate lots of hands and save them as text files so I could write a program to analyze them.

[inquiry]

ArcLight, on Sep 22 2004, 02:59 PM, said:

What is bridgelab.exe?

I looked for it on the web and couldn't find it.
I would find it useful if there was some software I could use to generate lots of hands and save them as text files so I could write a program to analyze them.

I think it would be much, much harder to write a program to analyze dealt hands than to write one to generate random or constrained hands (spoken by someone who doesn't program anything more complicated that a VCR).

To find bridgelab.exe on the web, just look up "Bridge Lab". They take the space out of the name because it is dos0based program. Here is one link

http://www.trsteiner.../en/tools.shtml

But if you want to set up hand constraints and deal them out, and then analyze what can and can't be made, I suggest Dealmaster pro. See this webpage

http://www.dealmaster.com/

Deal master is a little expensive given the other free tools, and mulit-function tools. But on the other hand, it does what you want it to do.

Be sure to look at the hand record (your generated hands) with "makable contracts".. I think this is what you might be looking for in one program, here is a linik to that feature as it is kind of hard to find on the pages

http://www.dealmaster.com/prthr1.htm

Ben

[ArcLight]

Ben, thanks for the links. I just played around with BridgeLab.exe. Its good for waht it does, and the other software on that page seems good as well.


> I think it would be much, much harder to write a program to analyze dealt hands than to write one to generate random or constrained hands (spoken by someone who doesn't program anything more complicated that a VCR).

I agree to a large extent. But one could write a program to look for borderline cases and flag them, and print out the evaluation of ZAR points, LTC, Rule of 20, etc. Just to compare.

And generating constrained hands might not be as simple as one may think. You can always brute force it, generating lots of hands, and discarding those that don't meet the criteria. But to generate a hand that meets your constraints without brute force requires some trade offs. Richard Pavlicek has an article on this:

http://www.rpbridge.net/7z71.htm

[hrothgar]

ArcLight, on Sep 22 2004, 09:59 PM, said:

What is bridgelab.exe?

I looked for it on the web and couldn't find it.
I would find it useful if there was some software I could use to generate lots of hands and save them as text files so I could write a program to analyze them.

You can't beat Dealer, originally written by Hans van Staveren

http://www.dombo.org/henk/dealer.html

[Syl20]

Ok.
I'am also sure we don't apply corrections the same way.
Here is my "algorithm":
(1) North is the one with 5+♠, 12-17Hcp
(2) I calculate raw Zar points for both hands
(3) I correct for trump honour(s) holding only for the South hand:
one for trump 10 or higher (maximum 2 points).
(4) I correct for trump length for both North (over 5) and South (over 3), allowing 3, 2 or 1 points per additional trump for void, singleton or doubleton.
(5) I correct for honours in short suits (in my arbitrary choice and we surely differ here):
Jack doubleton or singleton = 0
Queen doubleton or singleton = 1
King singleton = 2 (instead of 4)
King doubleton remains unchanged
Here are the results computed on 18000 boards:

Quote

Tricks        % Boards      ZarCorr
5            0,0005        46,2
6            0,0052        48,15385
7            0,0342        48,93859
8            0,0971        51,1359423274974
9            0,2009        53,5435540069686
10            0,2644        56,499621785174
11            0,2273        59,9111306643203
12            0,1273        63,509033778476
13            0,0431        68,04176
ZarCorr      %              Tricks
42            0,0003        7,33333333333333
43            0,0013        6,61538461538462
44            0,0017        6,88235294117647
45            0,0051        7,49019607843137
46            0,0088        7,54545454545455
47            0,0152        7,85526315789474
48            0,0182        8,14285714285714
49            0,0279        8,28673835125448
50            0,0374        8,52673796791444
51            0,0437        8,87871853546911
52            0,0549        9,04189435336976
53            0,0622        9,2491961414791
54            0,0673        9,44873699851412
55            0,0692        9,6893063583815
56            0,0681        9,92511013215859
57            0,0639        10,1533646322379
58            0,0646        10,3452012383901
59            0,062        10,5612903225806
60            0,0551        10,7186932849365
61            0,0483        10,9606625258799
62            0,0439        11,0728929384966
63            0,0365        11,2931506849315
64            0,0308        11,4318181818182
65            0,0245        11,6367346938776
66            0,0224        11,8080357142857
67            0,0151        11,8211920529801
68            0,0149        11,9395973154362
69            0,0108        12,1759259259259
70            0,0082        12,3292682926829
71            0,0038        12,4210526315789
72            0,0049        12,4081632653061
73            0,0027        12,6666666666667
74            0,0018        12,8333333333333
75            0,0013        12,6923076923077
76            0,0007        13                         

As you asked me, 12 boards do follow...

[Syl20]

... here are some boards

Quote

[Board "1"]
        S AJ6543
        H Q82
        D K74
        C K
S T            S 982
H JT75          H A94
D 932          D T5
C AQT87        C J6542
        S KQ7
        H K63
        D AQJ86
        C 93

[Board "2"]
        S AKQ52
        H AK8
        D 642
        C 97
S 96            S JT8
H J742          H QT5
D J9753        D KT8
C T5            C KJ63
        S 743
        H 963
        D AQ
        C AQ842

[Board "3"]
        S AJ852
        H A832
        D A62
        C Q
S Q            S K763
H KJ7          H 4
D QJ9854        D KT7
C T83          C 97652
        S T94
        H QT965
        D 3
        C AKJ4

[Board "4"]
        S AT9742
        H Q3
        D KQJ
        C KT
S Q6            S 5
H KJ8762        H AT954
D T952          D A864
C Q            C 632
        S KJ83
        H
        D 73
        C AJ98754

[Board "5"]
        S K86532
        H AQJ5
        D K
        C 53
S J            S Q7
H T9762        H 43
D Q42          D AJ8753
C KQT2          C 864
        S AT94
        H K8
        D T96
        C AJ97

[Board "6"]
        S Q9865
        H AKQ6
        D JT
        C T6
S A            S T72
H T75          H J98
D A8765        D KQ932
C J982          C K3
        S KJ43
        H 432
        D 4
        C AQ754

[Board "7"]
        S AJT76
        H Q96
        D 83
        C KQ6
S 82            S 94
H AJ            H K87432
D AJT6542      D 9
C J9            C T752
        S KQ53
        H T5
        D KQ7
        C A843

[Board "8"]
        S AK985
        H AJT7
        D 6
        C T42
S 642          S 73
H K93          H Q84
D Q742          D AK853
C A97          C K86
        S QJT
        H 652
        D JT9
        C QJ53

[Board "9"]
        S AK9862
        H AJ6
        D 87
        C Q2
S T4            S 5
H Q9732        H K84
D J43          D KQ62
C J93          C K7654
        S QJ73
        H T5
        D AT95
        C AT8

[Board "10"]
        S AK764
        H K
        D JT864
        C KJ
S 2            S Q3
H 765          H AJT932
D 972          D A3
C A97653        C T42
        S JT985
        H Q84
        D KQ5
        C Q8


[Board "5000"]
        S AQ976
        H J5
        D AQ8
        C JT7
S 84            S 3
H AQT732        H K94
D 652          D J973
C 52            C AKQ64
        S KJT52
        H 86
        D KT4
        C 983

[Board "10000"]
        S K9832
        H A4
        D AQ8
        C K54
S AT7          S 64
H J95          H KT2
D JT7532        D 94
C 3            C AQJ872
        S QJ5
        H Q8763
        D K6
        C T96

... and the results from "gib.exe":
Board#        Tricks        Zar      Zar(North)      Zar(South)    ZarCorr      Corr
1            11            61            31            30            63            2
2            12            60            33            27            59          -1
3            11            60            34            26            60            0
4            12            62            32            30            67            5
5            11            59            32            27            62            3
6            11            53            27            26            56            3
7            9              54            26            28            57            3
8            9              45            30            15            47            2
9            11            57            32            25            60            3
10          10            56            34            22            56            0
5000        8              49            29            20            52            3
10000        8              53            33            20            55            2

Hoping to converge soon with you.   

[Syl20]

Sorry i don't know how to handle the format for boards.

The order is North and the four last are South's hand. In the middle are East and West.
Simpler would be that i send you the original files if you wish ...